This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcc compile-time statistics
- From: Geoff Keating <geoffk at geoffk dot org>
- To: dewar at gnat dot com
- Cc: gcc at gcc dot gnu dot org
- Date: Sat, 18 May 2002 20:18:55 -0700
- Subject: Re: gcc compile-time statistics
- References: <20020519021309.70C51F28EB@nile.gnat.com>
- Reply-to: Geoff Keating <geoffk at redhat dot com>
> Date: Sat, 18 May 2002 22:13:09 -0400 (EDT)
> From: dewar@gnat.com (Robert Dewar)
> What do you think is causing the increase in time at -O0 (this
> actually is the most significant compilation time for us, since
> our customers do almost all their development at -O0 -- because
> of the debugging issues)?
As I said, I think it's mostly caused by extra RTL complexity, at
least between 3.0 and 3.1.
> It would be interesting to include 2.8.1 in these figures, because
> we have the impression that 2.95.3 already had incurred quite a
> slow down cost.
Sure, although I don't think it'll be very useful; it's too hard to
see what to do to the current compiler to get the time back, since the
current code is so different.
2.8.1 2.95.3 3.0.4 3.1
tree.c -O0 -g 0.49 0.59 0.72 0.88
tree.c -O1 -g 0.67 0.87 1.06 1.40
tree.c -O2 -g 0.93 1.37 1.71 2.23
--
- Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>